A journey through rhythms: from heartbeats to horology to the heart of AI.
Hello, I’m Kevin—sometimes also known as Dr. Abstracto!
If we were to meet at a café, our talk might quickly spiral into rhythms: the steady pulse of a heartbeat, the tick of a watch escapement, or the hidden cadence inside an AI’s attention mechanism.
This is the thread of my life: I call it Geofinitism. It is my philosophy where I strip away infinite abstractions, and explore my measurements of the word—shaped by rhythm, measurements, and interaction. Whether in medicine, horology, or physics, my curiosity has always returned to the same question: How do things hold together—and what rhythm guides their unfolding?
I’ve carried this inquiry through many guises: clinical scientist, engineer, horologist, writer, artist, cyclist. To me, these are not separate boxes, but all facets of one trajectory, one journey. And now, through Finite Mechanics, I’m seeking to frame physics and AI, not as infinities to be tamed, but as finite systems of interaction, measurements, and rhythm.
My journey began not in a traditional university, but at Norwich City Technical College studying Electronics. This practical foundation led me to a unique degree in Biomedical Electronics at Salford University. This was the only one of its kind in the country at the time, way back in 1981. It was an intense fusion of disciplines: medical physics, human physiology, cell biology, materials science, and core electronic engineering. Surrounded by just 20 highly selected students, I felt out of my depth at first: it was a profoundly formative time.
My first professional role was as an Installation Engineer for VG Analytical. My canvas was the world, and my brush was a toolkit for installing massive, multi-sector organic mass spectrometers, instruments like the VG ZAB shown below, which represented the absolute pinnacle of applied physics.
My role was to integrate the DEC-based computer systems that controlled these beasts. To do this, I had to understand everything: from the high-precision mechanics held in a vacuum to the component-level electronics. I would arrive at a lab in a company like DuPont or a major university, where the machine could already hold a vacuum and detect a beam, and my job was to make it sing to its basic specifications.
To this day, a VG Analytical mass spectrometer is displayed at the entrance of the Manchester Museum of Science and Industry. It’s the first thing you see after getting your guide book from the entrance kiosk. When visiting the museum, seeing it always reminds me of being 23 years old, installing million-dollar instruments that felt like they were from the future.
Although I loved the technology and the overseas travel, the constant time away from home was unsustainable. This led me to a 25-year career in medical engineering at Manchester Royal Infirmary where I could build a career based on my first degree in Biomedical Electronics.
With over twenty five years in Medical Engineering I worked on every kind of medical equipment imaginable. This ranged from measurements to therapy, and even surgical robots. Every moment was about safety and patient care. Medical equipment has to be safe and it has to work when its needed. Managing this equipment across one of the largest Healthcare Trusts in the UK was truly a daily challenge.
Slowly while working on all the different types of medical equipment, I built my research profile. My early research involved developing a system combining physiological measurements with fluoroscopy. Fluoroscopy is where you see those moving image in X-rays, and the measurements, the signals - they were the pressures created in the oesophagus when you swallow. The developed system became a clinical test used for over a decade. This work earned me an MSc and then building on this I gained my PhD, both from the University of Manchester. Towards the end of this research I was focusing on nonlinear analysis and neural networks for complex time-series data. This helped me see the connections of how nonlinear theories apply to AI and LLMs.
Over time, the work changed. I developed a complete software system for equipment management for over five of the UKs largest hospitals. My focus shifted towards management and administration. Sadly, the space for my own research slowly shrunk. Continuing changes within the Department began to make it difficult to have the time to develop new ideas and equipment. Eventually, I decided it was time to leave full-time work and seek a new kind of workshop.
I turned my full attention to horology—the art and science of watchmaking. I immersed myself in restoring high-grade Swiss watches. I became fascinated by a distinctive style from the late 1950s: electronic watches that used a transistor to drive a mechanical tuning fork, which in turn drove the hands via a mechanism of microscopic levers and jewels. It was miracle of micro-engineering and I became a master at repairing and restoring these fine timepieces.
This fascination grew into a multi-year project: to design and develop an entirely mechanical escapement that utilised a tuning fork. While patents for similar ideas existed, none had been successfully implemented. My approach was novel, using a non-contact method based on magnetic effects. I even filed a patent for the technique, and later deliberately allowed it to lapse, choosing to place the knowledge into the public domain. This work was a deep, tangible exploration of rhythm, resonance, and magnetic fields.
Completing the escapement project was not an end. It was a catalyst. The principles that guided it such as magnetism, fields, and precision measurements, sparked a deeper curiosity. I began to reconnect these tangible effects to the foundational physics I had studied decades earlier, like Schrödinger's wave equation, which had always felt abstract. Time was now my friend, I had time to think and connect ideas together.
I then began to ask a simple question: what if we could frame all of physical reality not with abstract infinities, but with finite, measurable axioms?
This question ignited the beginning of Finite Mechanics. It started informally with notes, sketches, and thought experiments. I wasn’t sure if creating even one equation was possible in this imagined world of finite axiom. I began re-examining Newton’s laws and other principles through this finite lens, exploring its implications for modern puzzles like galaxy rotation curves ("dark matter") and the perihelion of Mercury.
I started using LLMs to help writing software and a new journey started into older ideas: philosophy, how LLMs worked and how we find meaning. How does language work? how does mathematics really work? How are people and LLMs able to read my word and listen to me speak and get meaning? With all these questions in mind and then bringing all my previous experience on neural networks AI, signal analysis and nonlinear dynamical system theory, I went to work once again.
My journey is a work in progress, a personal pursuit driven by the same curiosity that has guided me from mass spectrometers to medical imaging to mechanical escapements. My work is not about grand claims; it’s simply about the process of seeking a more tangible understanding of our universe for myself. It is mine but I like the idea of sharing the path and leaving a faint impression of my footsteps behind. So join me on my journey every now and then and dip into my words and writing.
- Kevin
Core Expertise: Nonlinear Dynamics, Finite Axiomatic Systems, AI Interpretability, Interdisciplinary Research (Physics/Medicine/Design)
Links: finitemechanics.com | kevinhaylett.substack.com | GitHub
This list, drawn from his professional background, showcases a career of research at the intersection of medical engineering, physiology, and data science. This work demonstrates a consistent focus on developing novel analytical methods and instrumentation to understand complex systems. Note: This list is not exhaustive and has been selected to represent the scope and innovation, which provided the foundational experience for my current work on Finite Mechanics and Non-liner Dynamics in AI and Large Language Models.
Explore More: